When webpages become webapplications... the influence on statistics.


When web pages become web applications. Then page reloads are history and you interact with your webapplicationin your browser in the page and with AJAX/AHAH you will get the data in and out of the page. Nothing new, that is what the marketeers label web 2.0.

But when users interact with pages without having to reload ... when pages become application .. this also means that traditional ways of measuring user activity that are hosted and used as a service, such as Google Analytics, will not be able to tell you what a user has done on a pageapplication.

There used to be a time when the success of a website was measured in "hits", way back in the nineties. Then -due to the fact webpages consisted of many HTML elements like Cascading Style Sheets- the success was measured in pageviews. For the last couple of years, the success is measured in unique visitors since advertisers are not that interested in serving the same ad for the 10th times to the same person. And now, due to AJAX, we have to find a new way of measuring the success of webapplication.

Yahoo! understands this and has been offering Yahoo! User Interface services for some time now for free that can be used by a webmaster to give a more rich feeling towards the user. This way, Yahoo still gets to see who is doing what on a website and can offer this as a Analytics competitor as well as use the data for having a better advertising offering with behavioral targeting. The downside is however that to make your pages application, you do more then some icing on the cake; User Interface gadgets are nice. But the real deal is in enriching your data for webapplication use, not just your user interface.

AJAX-As-A-Service (as a new way of Server Based Computing) suffers the same problems as the old way of doing Server based Computing like Citrix has; looking to the world through a straw. Local data and terminal screens from remote do not mix well; a document that is saved on a local harddisk and accessed via a terminal service application still need lots of bandwidth and leads to high latency. The same kind of poblems you encounter when you use hosted AJAX service that are not integrated with your site; the (meta)data of the user is on the webserver and the AJAX application is served from another webserver not able to access all rich metadata.

Therefor only websites that are based on a Content Management Systems that is old will use these kind of services, modern CMS-es like Drupal ship with their own plug-able extend-able AJAX library. And if that CMS enables you to turn a web-page into a webapplication, you have two options to have the right statistics based on the right data; logging from the CMS or raw logging from your website. Drupal never has been good in providing logging analyses; it is good for technical webmasters but the analyses is not usefull at all for marketeers and managers. Analysing raw logging yourself is IMHO still the best way; you also get to see those firefox users that block urchin or have the best firefox plugin aivalble installed. However, it takes time and most of the reporting is by far not as fancy as Google Analytics.

So we will see how AJAX will influence web-analytics, maybe CMS-es will provide better statistics?

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Most web-analytics suck

The biggest problem for the ad sales folks is probably the inability to find what marks as a "new page", rather then the technology's inability. Sites that relay heavily on ad's usually have technology to counter this problem, however small sites using google ads probably will encounter this problem till Google makes a problem of it. I remember one (cant remember the name) company losing millions of ad revenue after a redesign of the site, because they had way less pages to navigate to content.

But when it comes to analysing user behaviour most web-analytics tend to suck a lot (the more expensive ones as WebTrends and HBX suck a bit less). What I think the overall solution to much of this is to have marketing people look at different metrics, because most google analytics don't really apply to web applications, but more to just ordinary content websites(they are working on changing that - like form abandonment). If you really know your way around google analytics there is some useful stuff for web applications, but it's still very slim.

I don't think CMS-es really have to provide better analytics, since for now there is no open-source solution that comes close to the detail that free and expensive ones offer. Of course, for the really geeky people who only trust raw apache logs you will never have a 'good' one.

When we get more Drupal sites that focus on e-commerce / ad powered sites we probably get a whole load of analytic modules.

google analytics beta

this also means that traditional ways of measuring user activity that are hosted and used as a service, such as Google Analytics, will not be able to tell you what a user has done on a page.

Actually, Google Analytics has a solution for this. Alas, it seems stuck in private beta.

Google has a non hosted solution which I hope gets integrated into Drupal.

Wow, you picked some great

Wow, you picked some great shots. but the first seem to be removed by flickr. i can see error message.

Post new comment


*

  • You may link to images on this site using a special syntax
  • You may post code using <code>...</code> (generic) or <?php ... ?> (highlighted PHP) tags.
  • Voting controls can be added to this post.
  • Lines and paragraphs break automatically.
  • Easily link to terms in various wikis. For help, see interwiki.
  • Allowed HTML tags: <A><I><LI><OL><U><UL><img><p><tt><table><hr><small><div><br><strike><b><pre><li><ul><td><tr><blockquote>
  • Insert Google Map macro. Create a macro